[chore] Fix KV cache block reuse flag name in quickstart_advanced (#3781)

Signed-off-by: Mike Iovine <[email protected]>

Author: mikeiovine

examples/pytorch/quickstart_advanced.py

@@ -66,9 +66,9 @@ def add_llm_args(parser):
 
     # KV cache
     parser.add_argument('--kv_cache_dtype', type=str, default='auto')
-    parser.add_argument('--kv_cache_enable_block_reuse',
-                        default=True,
-                        action='store_false')
+    parser.add_argument('--disable_kv_cache_reuse',
+                        default=False,
+                        action='store_true')
     parser.add_argument("--kv_cache_fraction", type=float, default=None)
 
     # Runtime
@@ -119,7 +119,7 @@ def setup_llm(args):
         enable_trtllm_decoder=args.enable_trtllm_decoder)
 
     kv_cache_config = KvCacheConfig(
-        enable_block_reuse=args.kv_cache_enable_block_reuse,
+        enable_block_reuse=not args.disable_kv_cache_reuse,
         free_gpu_memory_fraction=args.kv_cache_fraction,
     )
 

tests/integration/defs/test_e2e.py

@@ -1616,7 +1616,7 @@ def test_ptp_quickstart_advanced_eagle3(llm_root, llm_venv, model_name,
         f"{llm_models_root()}/{model_path}",
         "--eagle_model_dir",
         f"{llm_models_root()}/{eagle_model_path}",
-        "--kv_cache_enable_block_reuse",
+        "--disable_kv_cache_reuse",
     ])
 
 
@@ -1643,7 +1643,7 @@ def test_ptp_quickstart_advanced_deepseek_r1_8gpus(llm_root, llm_venv,
         "--kv_cache_fraction=0.95",
         "--max_batch_size=1",
         "--max_seq_len=3000",
-        "--kv_cache_enable_block_reuse",
+        "--disable_kv_cache_reuse",
     ])